Anthropic halted worldwide access to its newest cybersecurity AI models following a U.S. export-control directive that prohibits foreign nationals from using these tools anywhere in the world. This move transformed the company’s recent product launch into a flashpoint for federal security measures targeting advanced artificial intelligence technologies.
The order, issued by the U.S. government and grounded in national security concerns, came just days after Anthropic unveiled the Fable 5 and Mythos 5 models. The company initially sought to clarify what it considered a misunderstanding and aimed to restore access to users. However, it complied by disabling access for all international customers immediately.
The restriction has triggered significant backlash from cybersecurity leaders. A collective of 76 experts, including figures like Alex Stamos and Katie Moussouris, petitioned the White House to reverse the policy. The experts underscored that limiting access to these tools risks weakening cybersecurity defenses worldwide, as these AI models help identify software vulnerabilities that defenders rely on.
Major industry players also weighed in. Reuters reported that cybersecurity executives from companies such as Nvidia and Adobe urged the administration to reconsider the export-control stance. Meanwhile, Anthropic’s senior officials held meetings with government representatives in Washington, D.C., trying to negotiate a resolution amid the escalating dispute.
This episode signals a broader shift in how authorities regulate artificial intelligence. Traditionally focused on hardware like chips, U.S. export controls are increasingly applied directly to AI software models, leveraging access as a national security tool. The move illustrates the growing intersection between AI innovation, regulation, and geopolitics.
In addition to export restrictions, Anthropic has faced controversy over its AI’s military applications. The company declined to provide its models for fully autonomous weapons, resulting in Pentagon sanctions and blacklisting. Together, these developments place AI firms at the crossroads of security demands and governmental oversight, complicating their ability to operate globally and impacting their development strategies.